Re: xlink:href on simple element in DTD

Sigh.  This is why I shied away from putting a DTD in there for so 
long.  An argument could be made either way in this case, since there is 
normative prose in the spec about this (suggesting #IMPLIED is correct), 
but it describes what XLink applications *won't* do when href isn't there 
(suggesting #REQUIRED is correct).  I think I should probably keep it 
#REQUIRED and add some explanation.

BTW, I expect that a proposal on XLink infoset contributions will be made 
public soon, which should help a lot...

         Eve

At 08:53 AM 1/18/01 -0800, Jason Diamond wrote:
> > >Why does the DTD make the xlink:href attribute on the simple element
> > >REQUIRED, when 5.2 specifically allows this?  It makes the link
> > >untraversable, but so would an extended link with fewer than two
> > >resources, yet the DTD does not disallow this.
>
> > You're right.  It should be #IMPLIED.  I think I cut and pasted (blush).
>
>But the intro to the DTD says that "only constructs that have XLink-defined
>meaning are allowed." What's the XLink-defined meaning of a simple-type
>element with no href?
>
>Is the DTD trying to allow only "traversable" constructs? The comment below
>the location element seems to imply this--thus the reason for requiring href
>and label.
>
>Curiously, though, the label attribute is not required on resource. Are
>inline resources considered automatically traversable? Without a label, I
>don't see that it's possible to refer to a resource from an arc-type
>element.
>
>I understand that the DTD is non-normative but I'm trying to understand what
>an XLink-aware processor should be reporting to an application. An XLink
>infoset would be really nice.
>
>Jason Diamond
>http://injektilo.org/

--
Eve Maler                                          +1 781 442 3190
Sun Microsystems XML Technology Center    eve.maler @ east.sun.com

Received on Thursday, 18 January 2001 13:16:12 UTC